Professional Documents
Culture Documents
Modelling and Control of Liquid Level Sy
Modelling and Control of Liquid Level Sy
MARIZAN YAACOB
FK 2001 4
MODELLING AND CONTROL OF LIQUID LEVEL SYSTEM
By
MARIZAN YAACOB
October 2001
Abstract of thesis presented to the senate of Universiti Putra Malaysia in fulfilment
of the requirement for the degree of Master of Science
By
MARIZAN YAACOB
October 2001
Faculty: Engineering
Control system modelling is difficult and delicate process. In most cases, the model
cannot be obtained directly from the plant due to the complexity of the system
allow analytical formulation of equations. On the other hand, the model can be
data was obtained by applying a step input and measuring the output response of
the system. A transfer function was estimated by using curve fitting technique. This
estimated model has been refined by tuning the parameters e.g. gain, time constant,
Control design has also been carried out based on the model obtained through
ii
The same parameter settings of the controllers was applied to the real plant and
performance of the system under PD control was satisfactory. This also proved that
the model derived was a good estimate of the real plant and it can be used for
further control design.
iii
Abstrak tesis yang dikemukakan kepada Senat Universiti Putra Malaysia sebagai
memenuhi keperluan untuk ijazah Master Sains
Oleh
MARIZAN YAACOB
Oktober 2001
Fakulti: Kejuruteraan
Permodelan sistem kawalan adalah susah dan prosesnya adalah rumit. Dalam
kebanyakan kes, model tidak dapat diperolehi secara langsung daripada loji
Dalam pengajian ini, model bagi sistem aras cecair yang akan diterbitkan. Data
parameter seperti gandaan, pemalar masa, frekuensi tabii tak teredam dan nisbah
redaman sistem.
iv
Parameter yang sarna dari kedua-dua pengawal kemudiannya digunakan pada loji
v
ACKNOWLEDGEMENTS
supervisor Dr. Samsul Bahari Mohd. Noor, Prof Madya Ir. Dr. Norman Mariun and
Dr. Sinan Mahmod for their invaluable guidance, constructive suggestions and
Lastly, I would like to thank my husband and family for their never ending love,
patience and support which has enabled me to complete the project successfully.
vi
I certify that an Examination Committee met on 13th October 2001 to conduct the
final examination of Marizan Yaacob on her Master of Science thesis entitled
"Modelling and Control of Liquid Level System" in accordance with Universiti
Pertanian Malaysia (Higher Degree) Act 1980 and Universiti Pertanian Malaysia
(Higher Degree) regulation 1981. The Committee recommends that the candidate
be awarded the 'relevant degree. Members of the Examination Committees are as
follows
vii
This thesis submitted to the Senate of Universiti Putra Malaysia has been accepted
as fulfilment of the requirement for the degree of Master of Science
AINI IDERIS, Ph D
Professor,
Dean of Graduate School
Universiti Putra Malaysia
V111
DECLARATION
I hereby declare that the thesis is based on my original work except for quotations
and citations, which have been duly acknowledged. I also declare that it has not
been previously or concurrently submitted for any other degree at UPM or other
institutions.
( MARIZAN YAACOB)
ix
TABLE OF CONTENTS
Page
ABSTRACT ii
ABSTRAK iv
ACKNOWLEDGEMENTS VI
APPROVAL Vll
DECLARATION ix
LIST OF TABLES xii
LIST OF FIGURES xiii
LIST OF ABBREVIATIONS xvii
CHAPTER
1 INTRODUCTION 1
1.1 General Overview 1
1.2 Objective of This Work 3
2 LITERATURE REVIEW 5
2.1 System Identification 5
2.2 Obtaining Model from Experimental Data 7
2.3 Process Control 8
2.4 Linear Systems 9
2.5 Transient Response 10
2.6 Step Input 11
2.7 Errors 13
2.7.1 Disturbance Error 13
2.8 System Stability and Performance 16
2.8.1 Relative Stability 17
2.8.2 Instability 19
2.9 System Response with Zeros 20
2.10 Proportional, Derivative and Integral Controller 22
2.10.1 Selecting Controller on Existing Process Plant 23
2.11 Overdamped Second Order System 25
2.12 Curve Fitting Method for Analysing Second Order
Overdamped Response 31
2.13 Summary 34
3 METHODOLOGY 35
3.1 Introduction 35
3.2 Description of Fill Level Control System 37
3.3 Find Step Response of Actual System 40
3.3.1 Equipment Configuration 41
3.3.2 Software Configuration 42
3.4 Find the Estimated Model 45
3.4.1 Analysis of experimental data using curve
fit procedure 47
3.4.2 Applying PID controller to the system 50
x
4 RESULTS AND DISCUSSION 53
4. 1 Introduction 53
4.2 Experimental Response 54
4 .3 Analysis of Experimental Data using Curve Fit Procedure 59
4.3 . 1 Analysis of Data of Test 1 60
4 . 3 .2 Analysis of Data of Test 2 63
4 . 3 . 3 Analysis of Data of Test 3 66
4 . 3 .4 Analysis of Data of Test 4 69
4.4 Average Transfer Function 72
4.5 Transfer function modification 74
4. 5 . 1 Modification of zeros location 74
4.5.2 Effect o fDamping Factor ( S) and
Natural Frequency (�) to System Response 85
4.6 Model Validation and Identification the Controller Parameters 91
4.6. 1Proper Method for Setting PID Parameters 95
4.6.2 Adaptation ofPID Controller to the Estimated Model 98
4.6.3 Adaptation ofPD Controller to the Estimated Model 105
4.7 Summary 1 08
REFERENCES 1 15
APPENDICES
xi
LIST OF TABLES
Table Page
4. 1 Data of test 1 1 49
xii
LIST OF FIGURES
Figure Page
2.2 Step input signal (a) Practical step (b) Idealized unit step. 12
xiii
4.8 Step response of estimated transfer function Gl(s). 62
xiv
4.30 Location of poles and zeros of transfer function GS2(s) 80
xv
4.48 Step response of the system with PID controller. 99
XVI
LIST OF ABBREVIAnONS
X(s), W(s), Y( s), C(s) variables which have been transformed into the s domain.
r Time constant
t; Damping factor
xvii
CHAPTERl
INTRODUCTION
The project title is modelling and control of level control system. This
project is involved in controlling the water level in process control system. The
process control is applied using real system RT210 Process Control Rig, G.D.N.T
controllers (P, PI and PID controllers, two point and three-point controllers).
The approach of this project is to observe the behaviour of the system in its
modelling methods. The time domain is the natural way to look at the behaviour of
such systems since this is how to see the response of real systems. The
experimental data of the system is obtained in real time sampling. The data is
mathematical model of the system. The measured response is compared with the
computer aided analysis. The model with the closest fit can then be chosen to
1
Matlab program is an alternative approach to model the system in the form
physical harm. Indeed, modification can be carried out when a new setting
parameters is found to work successfully, it then can be tried out on the actual
system with a much greater degree of confidence. Hence, system perfonnance can
be investigated within and outside the expected operation range, thus any problem
system is concerned.
model of the system. The model estimation must be the closely approximated to the
real system. Controllers are cascade control devices and they are inserted into an
existing control loop to fonn part of forward gain of the control system. These
devices are widely used in process control system and are usually employed as a
part of the control circuitry on process control valves and hydraulic actuators.
Controller contain three tenns which are proportional, integral and differential is
referred to as PID controller. The coefficients of PID controller can be found using
have been finalised, the compensation tenn will be sorted out within this stage of
the analysis.
2
1.2 Objective of This Work
computer aided design of linear control system is concerned. This problem has been
attempted previously using pure numerical methods by Laub [2] and efficient
on the imaginary axis have been designed. Although these algorithms are useful in
many applications, analytical expressions for the transfer functions are often
desirable because they contain the most complete characterisation of linear time
invarient systems. They allow for the numerical evaluation of the transfer functions
and their derivatives at many points in the complex plane. Kailath [3] express that
The objective behind this project is to find the model which represent the
dynamic characteristics of the actual system. Chan [4] explained that the dynamic
On the other hand, experimental data from this system is readily available. Chotai
[S] and Rogers [6] had been developed a design technique based on experimental
data to circumvent the problem. Franklin [7] discussed several reasons for using
model derived from transient response data can be reliable for designing the control
this project in order to find the dynamic model of level control system. The
3
dynamic model represents the dynamic sensitivity of a system which contains all
1. to evaluate the output response of system and identify the system order,
equations,
5. to perform numerical analysis with the aid of the computer (calculation and
system behaviour,
and
4
CHAPTER 2
LITERATURE REVIEW
from data obtained from measurement on the data input and output. All these
procedures require that the order of transfer function be assumed. The methods may
pulse, or the response to a more general input. Generally, the methods can be
fitted to the available data in some optimal manner. These results in the "best" fit of
the transfer function to the data but not necessarily result in a good model. For
example, second order transfer function can be fitted to the input-output data of a
highly nonlinear physical system. The result is a best fit of the transfer function to
the data, but in this case the best fit is not good enough. Hence these methods, as
with all system modelling techniques, must be used with great care [9].
known transfer function, the response to a given forcing function can be evaluated.
transfer function, and to determine or confirm numerical values for the parameters,
5
or alternatively, for a so-called 'black box' system where governing equations are
subjecting the components to, say, a step change of input, recording the resultant
output response occurs later than the time of application of step change, then the
transfer function must be higher order than first. Attempt should be made to see
whether a second order transfer function would give a close fit to the response
can give a useful indication about linearity. The component can be considered
linear provided the shape of the response curve resulting from a step change of
output of the system will change as a result of specific types of input change. On
not the system behaviour is satisfactory and, if not an attempt made to improve the
response by realisable modification to the system. With practical systems the exact
form of the input excitation function may be known in advance, but most frequently
the input will vary in a somewhat random and hence largely unpredictable manner
(such would be the case where the ambient temperature is a significant input
6
2.2 Obtaining Model from Experimental Data
There are several reasons for using experimental data to obtain a model of
the dynamic system to be controlled. In the first place, the best theoretical model
as in the case of very rigid spacecraft, the theoretical model is extremely good.
metalworking, the theoretical model is very approximate. In every case, before the
final control design is done, it is important and prudent to verify the theoretical
physics of the process is poorly understood, the only reliable information on which
to base the control design is the experimental data. Finally, the system is sometimes
subject to on-line changes, which occur when the environment of the system
changing the control parameters. This requires model for a new conditions, and
experimental data are often the most effective, if not the only, information available
for the new model. There are four kinds of experimental data for getting the model
[7]:
2. Frequency response data, which results from exciting the system with